Gene expression signatures of adult T-cell leukemia: is treatment response prediction on the horizon?

In recent years, novel therapeutic approaches and, particularly, targeted treatment strategies have been derived from increasing insights into the molecular pathogenesis of cancer. At the same time, it has become increasingly clear that novel therapies may be variably efficacious in distinct subsets of molecularly defined tumor entities. Methodological developments in the post-genome era have opened new options to study complex effects of drugs in a genome-wide manner and, thus, have paved the way to explore intra- and interindividual tumor heterogeneity and to investigate the molecular networks involved in the response or resistance to therapy.

Specifically, gene expression profiling has allowed the identification of molecular tumor subtypes, and is being increasingly applied to study the molecular effects of specific treatment approaches or the prediction of response to a given therapeutic regimen. In hematological malignancies, for example, a consistent in vivo gene expression signature in response to treatment with fludarabine was identified in chronic lymphocytic leukemia (CLL), demonstrating an activation of p53 and its downstream targets. This specific signature was apparent in CLL cells that responded to fludarabine treatment, but not in fludarabine-resistant CLL cells [1]. This observation is in line with reports showing that patients with CLL with inactivation of p53 are more likely to have fludarabine-refractory disease, whereas they might respond well to anti-CD52 antibody therapy (alemtuzumab) [2–4].

In diffuse large B-cell lymphoma (DLBCL), gene expression profiling has led to the identification of two major molecular subgroups, namely the germinal center B-cell (GCB)-like and the activated B-cell (ABC)-like subtypes, the latter being characterized by constitutive activation of the nuclear factor κB (NFκB) pathway and inferior survival times following therapy with cyclophsphamide, doxorubicin, vincristine, prednisone (CHOP)-based [5] or CHOP plus rituximab (R-CHOP) [6] regimens. Very recently, Dunleavy and colleagues demonstrated elegantly in 49 patients with relapsed DLBCL that treatment with bortezomib, a proteasome inhibitor which acts at least partially via inhibition of NFκB signaling, combined with chemotherapy is more effective in patients with relapsed ABC-type DLBCL [7]. These studies highlight the value of genome-wide analyses aiming at the molecular definition of tumor subgroups with different biological features that may be reflected in their clinical course and their distinct response to specific therapies.

Ideally, such studies should be performed in the context of clinical trials or, at least, in patient cohorts that received a homogeneous treatment. This approach was followed in a study reported by Alizadeh and colleagues in this issue of Leukemia and Lymphoma [8]. The authors performed gene expression profiling in a series of patients with adult T-cell leukemia at the time of diagnosis and 24 h after treatment initiation with a combination of zidovudine (AZT) and interferon α (IFNα), in order to identify molecular features of the tumor cells in responding versus non-responding patients.

Adult T-cell leukemia/lymphoma (ATLL) is a distinct malignant T-cell lymphoproliferative disorder that is causally linked to infections with the human T-cell lymphotropic virus type 1 (HTLV1), a retrovirus which is endemic in southwestern Japan, the Caribbean basin, and parts of central Africa and South America. Accordingly, ATLL constitutes up to 25% of all T-cell lymphomas in these areas, whereas it is a rare disease in North America and Europe [9,10]. The exact mechanisms of oncogenic transformation by HTLV1 remain to be elucidated, but there is evidence that the viral Tax protein is involved in the initiation of oncogenic transformation [11]. However, only a minority of individuals infected with HTLV1 develop ATLL, usually after a long latency period, indicating that the HTLV1 infection alone might be insufficient for full malignant transformation and that additional factors, such as genetic alterations accumulating over time in the infected cells, may be involved in this process.

Based on the presentation and the clinical course, four subtypes of ATLL can be discerned, including the acute and the lymphoma types, both of which show an aggressive clinical behavior with median survival times of less than 1 year, whereas the chronic and the smoldering subtypes are more indolent [11,12]. The current treatment approaches of ATLL range from watchful waiting in the indolent forms to multimodal strategies including chemotherapy, antiviral therapy, allogeneic stem-cell transplant, and molecular targeted treatment approaches in the clinically challenging aggressive subtypes [12]. In view of the very poor outcome of the latter despite intensive therapeutic efforts, a better understanding of the molecular basis of the disease and novel therapeutic approaches are urgently needed. While some promising results have been reported in ATLL using a combination therapy of AZT and IFNα [13–15], these studies do not offer detailed insights into the underlying molecular mechanisms of response to therapy. Alizadeh's study [8] that reports molecular profiling results of ATLL cells investigated before and after combination therapy with AZT and IFNα sheds some light on this issue. Although the number of patients that were included in the study was quite limited, the authors were able to identify specific gene expression signatures that are associated with the responsiveness of the tumor cells to this therapy. Specifically, treatment with AZT/IFNα induced a marked up-regulation of interferon response genes in vivo, whereas cell cycle-associated genes were silenced. Interestingly, patients with ATLL not responding to AZT/IFNα failed to show the interferon response signature. Moreover, genes that were anomalously overexpressed in ATLL could be identified, such as the transcriptional regulator LMO2 that may be of relevance in the pathogenesis of ATLL.

Overall, Alizadeh's study highlights the enormous potential of modern genomic approaches to determine molecularly defined tumor subsets that will or will not respond to a given therapeutic approach. The application of these molecular techniques in the context of homogeneously treated patient cohorts or clinical trials will ultimately help to select the best choice of treatment for each individual patient.